Axial Spondyloarthritis: Mimics and Pitfalls of Imaging Assessment

Axial Spondyloarthritis: Mimics and Pitfalls of Imaging Assessmentdoi: 10.3389/fmed.2021.658538
Frontiers in Medicine | www.frontiersin.org 1 April 2021 | Volume 8 | Article 658538
Edited by:
Xenofon Baraliakos,
*Correspondence:
Rheumatology,
Frontiers in Medicine
Machado PM (2021) Axial
Spondyloarthritis: Mimics and Pitfalls
Axial Spondyloarthritis: Mimics and Pitfalls of Imaging Assessment
António Proença Caetano 1†, Vasco V. Mascarenhas 2,3† and Pedro M. Machado 4,5,6*†
1 Radiology Department, Hospital de Curry Cabral, Centro Hospitalar Universitário Lisboa Central, Lisbon, Portugal, 2Musculoskeletal Imaging Unit, Grupo Luz Saúde, Radiology Department, Imaging Center, Hospital da Luz, Lisbon, Portugal, 3 EpiDoC Unit, Chronic Diseases Research Centre, NOVA Medical School, Lisbon, Portugal, 4Centre for Rheumatology &
Department of Neuromuscular Diseases, University College London, London, United Kingdom, 5National Institute for Health
Research (NIHR) Biomedical Research Centre, University College London Hospitals National Health Service Foundation
Trust, London, United Kingdom, 6Department of Rheumatology, London North West University Healthcare National Health
Service Trust, London, United Kingdom
Axial spondyloarthritis (axSpA) is a chronic inflammatory disorder that predominantly
involves the axial skeleton. Imaging findings of axSpA can be divided into active
changes, which include bone marrow edema, synovitis, enthesitis, capsulitis, and
intra-articular effusion, and structural changes, which include erosions, sclerosis, bone
fatty infiltration, fat deposition in an erosion cavity, and bone bridging or ankylosis. The
ability to distinguish between imaging lesions suggestive of axSpA and artifacts or lesions
suggestive of other disorders is critical for the accurate diagnosis of axSpA. Diagnosis
may be challenging, particularly in early-stage disease and magnetic resonance imaging
(MRI) plays a key role in the detection of subtle or inflammatory changes. MRI also
allows the detection of structural changes in the subchondral bone marrow that are not
visible on conventional radiography and is of prognostic and monitoring value. However,
bone structural changes are more accurately depicted using computed tomography.
Conventional radiography, on the other hand, has limitations, but it is easily accessible
and may provide insight on gross changes as well as rule out other pathological features
of the axial skeleton. This review outlines the imaging evaluation of axSpA with a focus
on imaging mimics and potential pitfalls when assessing the axial skeleton.
Keywords: axial spondyloarthritis, magnetic resonance imaging, radiography, computed tomography, differential
diagnosis, pitfall, normal variant, mimic
INTRODUCTION
Axial spondyloarthritis (axSpA) is an umbrella term encompassing a group of chronic immune- mediated inflammatory diseases of the axial skeleton. This group includes patients with radiographic axSpA, with established sacroiliitis on radiographs, and a further subgroup called non-radiographic axSpA, who typically have evidence of sacroiliitis onmagnetic resonance imaging (MRI) in the absence of definite radiographic changes.
Historically, the diagnosis of axSpA has often been delayed since radiographic abnormalities may take years to develop. Computed tomography (CT) allows for detection of smaller structural lesions in patients with chronic sacroiliitis that would otherwise be invisible on conventional radiography, thus aiding in the diagnostic work up of axSpA. In recent years, the introduction of MRI into clinical practice has facilitated earlier diagnosis of axSpA, and therefore earlier initiation of appropriate treatment. The Assessment of Spondyloarthritis International Society (ASAS) MRI working group has recently generated a consensus update on standardized definitions for MRI
lesions in the sacroiliac joint (SIJ) of patients with axSpA (1). Multi-reader validation performed by the working group demonstrated substantial reliability for the most frequently detected lesions and comparable reliability between active and structural lesions. A similar exercise has been conducted for spine lesions and recently published in abstract format (2). The new consensus definitions for MRI lesions in the spine will replace a previous consensus manuscript by the same group (3).
Importantly, the full range and combination of active and structural lesions of the SIJ and spine should be taken into account when deciding if the MRI scan is suggestive of axSpA or not (i.e., contextual interpretation of active and structural lesions is key to enhancing diagnostic utility of MRI in patients with suspected axSpA), as imaging cannot be viewed in isolation and needs to be interpreted in the light of clinical presentation and results of laboratory investigations (4, 5).
MRI evaluation of the SIJ can be quite challenging even for experienced radiologists, due to several pitfalls. Being familiar with the main imaging findings and terminology of axSpA (Table 1) as well as knowing the topographic distribution of common and uncommon conditions involving the SIJ is key to establishing a confident diagnosis (Figures 1A,B).
In this article, we will review common and uncommon pitfalls, congenital disorders, normal variants and pathological
TABLE 1 | Imaging findings of active and chronic changes of the sacroiliac joint
and spine in axial spondyloarthritis.
Sacroiliac joint Spine
erosion
“backfill”)
• Osteopenia
*The terms “Romanus spondylitis” and “shiny corners” have been used in the context
of MRI assessment but should be avoided as they were initially described in plain
radiographs: “Romanus spondylitis” appears as irregularity and erosion involving the
anterior and posterior corners/edges of the vertebral endplates, while “shiny corners”
represent reactive sclerosis secondary to inflammatory process.
conditions that may mimic spondyloarthritis affecting the axial skeleton.
ANATOMY OF THE SACROILIAC JOINTS AND THE SPINE
The SIJ is the largest joint of the axial skeleton and consists of an amphiarthrosis, exhibits restricted mobility and is separated into a ligamentous (posterior) and synovial (anterior) component. The cartilage covering the synovial segment is thicker on the sacral side and, thus, less prone to lesions (6).
The SIJ is lined by a capsule. Several ligaments contribute to its stability and may be affected in axSpA, namely the anterior and posterior sacroiliac ligaments and interosseus ligament connecting the tuberosities of the sacrum and ilium deeply in the ligamentous portion. The intervertebral disc is also an amphiarthrosis and is comprised of an inner core—the nucleus pulposus—and an outer fibrous ring—the annulus fibrosus. There is also cartilage lining on the superior and inferior vertebral plates that protects the subchondral bone at this level. The inner core is generally spared in axSpA, but the annulus fibrosus attaches to the periphery of the vertebral plates where there is no cartilage protection, and interweaves with the anterior and posterior longitudinal ligaments of the spine, working as an enthesis.
Besides the annulus fibrosus, several ligaments stabilizing the spine are prone to inflammation at their insertion point, namely the anterior and posterior longitudinal, supraspinous, interspinous, intertransverse ligaments, and ligamentum flavum.
NORMAL VARIANTS AND PITFALLS
In this section we will describe potential anatomical variants and pitfalls of the SIJ and spine that may mimic axSpA findings (Table 2).
Coil Effect Technical artifact responsible for artificial hyperintensity of structures near the receiver coils. These structures may be mistaken for bone marrow or soft tissue edema. Such findings, however, can be distinguished from true inflammatory changes due to their topographic distribution, which is predominantly peri-articular in the latter scenario.
Inadequate fat suppression is higher in patients with higher body mass index (BMI). Radial k-space sampling, an imaging reconstruction technique utilized in MRI data acquisition that is relatively insensitive to motion artifacts, seems to have a positive impact on image quality in such patients (7). Another solution might be to change saturation techniques, from a spectral pre- saturation of fat signal to a short tau inversion recovery (STIR) sequence, which homogeneously suppresses fat, with the caveat of reducing overall signal.
The type of coil also seems to have a significant impact on image quality, more so when combined with the correct sequence in reducing artifacts. This combination yields the best inter- observer agreement for bone marrow edema (BME) detection and lowest number of doubtful BME zones (8).
FIGURE 1 | Imaging of the sacroiliac joint—Topographic distribution of main anatomical variants and pathological conditions that mimic axSpA, separated by
quadrants of each articular surface (A) and orthogonal planes (B), namely coronal oblique (right upper image) and axial oblique (right lower image).
Phase-Encoded Motion Artifacts Motion artifact may occur due to vessels, intestinal motion and patient motion. This artifact may cause blurring or a hyperintense image superimposed at or adjacent to the SIJ and mimic BME. Cross-reference between two perpendicular planes may allow avoidance of overcalling lesions (8, 9). Cerebrospinal fluid (CSF) and blood motion artifacts are also common in spine MR imaging (10).
Again, radial k-space sampling offers a higher signal-to-noise ratio and contributes to reduction in motion-related blurring. Application of motion-resistant sequences is also recommended (7). Other techniques may be employed, such as increasing the number of excitations, changing the phase-encoding axis (along the direction of CSF flow) or applying pre-saturation pulses outside the region of interest (11). Repetitive motion from breathing or cardiac motion may be reduced with gating techniques that perform data acquisition at specific intervals.
Blood Vessels Blood vessels coursing close to the SIJ and spine along the acquisition plane may simulate bone marrow or soft tissue edema, synovitis or joint fluid on fluid-sensitive sequences. They present as linear hyperintensities along the acquisition plane and may ramify with other vessels on adjacent slices. CSFmotionmay also be an issue in spine imaging.
Intense vascularization may be seen at the transition between cartilaginous and ligamentous portion of the joint, at the
ligamentous portion and adjacent to certain anatomical variants such as the iliosacral complex and the semi-circular sacral defect, which are described in more detail below (12). Vessels can also run along bones.
Normal Marrow Changes Red marrow replacement occurs in a centrifugal fashion in individual bones and in a centripetal fashion in the skeleton. The extremities are primarily affected by this physiological aging phenomenon and, by the middle of the third decade (13), most of the bone marrow in long bones has an overall fatty marrow. Individually, conversion into fatty marrow starts in the diaphysis of long bones and progresses to the metaphysis, ultimately converting the distal epiphysis and, lastly, the proximal epiphysis. In the axial skeleton, the pattern of reconversion is less predictable and several patterns have been described in the spine (14). In the pelvis, small pockets of yellow marrow arise in the third decade in the acetabulum and anterior ilium. In the sacrum of male patients there is higher fat content in the lateral masses compared to females (15) and localized aggregates of fat marrow in the lumbar spines and lateral sacral ala are considered normal variants (16).
General Population and People With Chronic Non-specific Back Pain Weber et al. showed that 25% of healthy individuals have signs suggestive of sacroiliitis on MRI (17). Similarly, Arnbak et al.
Caetano et al. Imaging of Axial Spondyloarthritis
TABLE 2 | Congenital disorders and normal variants of the sacroiliac joints and spine that mimic axial spondyloarthritis.
Condition Type Characteristic features
Blood vessels – Location–ligamentous portion of the SIJ, adjacent, adjacent to anatomical variants,
lower ilium (partial volume)
Low SPARCC scores
Sports/exercise related – Topographic distribution overlaps with axSpA
Port-partum – Extent and distribution indistinguishable from axSpA
Structural changes are rare
Schmorl nodes – Location–lower thoracic and upper lumbar vertebrae, along the nucleus pulposus axis
Block vertebra Congenital Location–cervical segments
Other associated conditions
Other–women
Other–women
Other–triangular shape
Other–unilateral, women
Accessory iliac joints Location–between iliac and sacral surfaces at posterior joint
Semicircular defect articular surface Location–ligamentous portion, postero-superior, focal sacral depression
Other–women, bilateral
Transitional vertebrae/Bertolotti
Types II and IV correlate with symptoms and disc herniation
Spina bifida occulta – Location−5th lumbar segment
Other–correlation with spondylolysis
Intra-osseous pneumatocyst – Location–iliac bone adjacent to SIJ, lumbar or cervical spine
Tarlov cysts – Location–sacrum
Other–bilateral, women, 40 years-old
found that, in 1,020 unselected individuals, 21% had sacroiliitis on MRI according to ASAS criteria.
Other authors (18) suggested that one fourth of asymptomatic individuals and more than half of women with post-partum back pain without axSpA had MRI positive sacroiliitis according to ASAS criteria. This study also showed that frequent runners have similar findings compared to asymptomatic individuals and that scoring high on a specific scoring system used for axSpA activity (Spondyloarthritis Research Consortium of Canada Scoring System for Sacroiliitis, SPARCC) is rare in healthy individuals and runners. Furthermore, deep lesions are specific for axSpA-related sacroiliitis and BME lesions in healthy individuals are preferentially located in the lower iliac bone.
Indeed, others studies have documented the presence of BME in healthy individuals without any symptoms of low back pain, which does not change in the setting of mechanical stresses or physical exercise (19) (Figures 2A,B). Recently, however, a large population study by Baraliakos et al. (20) confirmed a high prevalence of inflammatory and fatty changes in the SIJ and spine, which increases in frequency with age, suggesting a mechanical factor to their development.
Sports/Exercise Related BME Evaluation of MRI lesions in athletes poses a significant challenge when attempting to discriminate healthy individuals from early axSpA. In fact, 30–35% of recreational runners and 41% of elite hockey skaters have shown ASAS criteria for axSpA when evaluated for sacroiliitis (17). Partial volume effect of vascular structures, mechanically triggered BME due to axial strain and normal anatomical variants are thought to be the main reasons for such findings. Applying a complementary semi-axial plane for evaluation seems to significantly reduce ASAS positivity to 20 and 18%, respectively for recreational runners and elite hockey skaters (8).
The two most common portions of the joint affected by BME are the anterior upper sacrum and the posterior lower ilium, the latter associated with partial volume effect of vessels and deep iliac ligament insertion. Unfortunately, it is well-recognized that the early incipient findings of inflammatory changes in axSpA patients show a topographic overlap with BME associated with constitutional features on the dorso-caudal portion of the SIJ, at the posterior lower ilium.
Low-grade BME lesions may indeed have several potential triggers such as mechanical overload or stress, anatomical
Caetano et al. Imaging of Axial Spondyloarthritis
FIGURE 2 | T1WI (A) and STIR image (B) of a military subject showing a small, peri-articular, area of bone edema (arrow) on the iliac side of the right sacroiliac joint.
T1WI (C,E) and STIR (D,F) images of a post-partum female with bilateral foci of bone edema (arrows) adjacent to the sacroiliac joint.
variations, heavy load work, overweight and post-partum. Discriminative factors that may indicate possible or probable axSpA have not been determined—BME extension alone has not proven to be a relevant criterion (9), but evaluation of
extent and topographical pattern might be able to reduce false-positive assessments of ASAS MRI positive sacroiliitis. Assessment of other structural features and active lesions may improve specificity (21, 22).
Post-partum Low back pain is common during pregnancy and shortly after birth, typically resolving 6 weeks post-partum. Some patients, however, experience long-standing low back pain more than 6 months after childbirth (23).
Causes for post-partum symptoms are multifactorial and involve mechanical stress and hormonal changes, child and birth characteristics (24). Post-partum SIJ infection is an important differential diagnosis as it accounts for 15% of septic sacroiliitis events; auto-inflammatory conditions may also manifest during pregnancy or after childbirth.
Agten et al. (25) compared the SIJ of post-partum women and women with known axSpA and found no distinguishable features based on extent and distribution, making it difficult to avoid overcalling axSpA in such patients (Figures 2C–F). Presence of structural changes, however, was more frequent in the axSpA group and only rarely found in the post-partum group. Furthermore, pain referral and pain intensity were not correlated with BME in the post-partum group. Importantly, puerperal diastasis of the pubic symphysis and SIJ is physiological to some degree and only in rare situations is associated with complications (26).
Nonetheless, Winter et al. showed positive findings on MRI of post-partum women with back pain, which was consistent with previous data that reported 60% of such patients having SIJ BME lesions on MRI (18, 27).
Schmorl Nodes Schmorl nodes correspond to herniation of nucleus material through the endplate of the vertebral bodies into the subchondral bone (28).
Schmorl nodes are usually marginated by a well-defined sclerotic border which may be irregular and are more prevalent in the lower thoracic and upper lumbar segments. The etiology of Schmorl nodes is multifactorial, involving trauma and congenital causes. There is also an association with smoking habits, vertebral body length, and age (28). Patients with Schmorl nodes may be asymptomatic or present with low back pain, and an association with degenerative spine disease and disc degeneration has been established (29). If Schmorl nodes become symptomatic, MRI may demonstrate inflammation and edema in the bone marrow surrounding the Schmorl node. Vertebroplasty has been tried out and proven to be effective and safe when symptoms do not resolve with medical or physical therapy (30, 31).
Acute Schmorl nodes may mimic other inflammatory conditions affecting the spine. Imaging features are of a concentric ring-type edema and involvement of the adjacent end-plate to the herniated node, without diffuse signal abnormalities (32).
Block Vertebrae Block vertebrae may be congenital or acquired. Congenital blocked vertebra is generally found in the cervical spine and associated with Klippel-Feil syndrome (short neck, low hair line, and neck movement restriction). Other abnormalities associated with congenital block vertebra include syringomyelia, diastematomyelia, or tethered cord (33).
Acquired vertebral fusion may be a desired surgical outcome in cases of advanced degenerative disc disease or cases of joint instability (34, 35). Also, late-onset ankylosing spondylitis with extensive calcification may lead to bamboo spine due to dystrophic and ligament calcifications so extensive that they merge both endplates of the disc joint. Interbody fusion requires disc removal through a posterior or anterior approach, insertion of a bone graft and/or fusion hardware. The purpose is to achieve an arthrodesis along the disc space. Complications include pseudarthrosis, when bone bridging does not develop or is insufficient. Studies to evaluate post-operative fusion include CT, MRI and bone scintigraphy.
SIJ Anatomical Variants Synovial recesses, bony and cartilage clefts that may mimic bone erosion, intense vascularization on the ligamentous portion that enhances avidly and fat infiltration of the sacral bone marrow without pathological significance may be evident on SIJ imaging and are addressed in other sections of this article.
In this section we briefly describe the seven anatomical variants of the SIJ that have been documented to date. The morphology of the sacral and iliac surfaces is well-depicted on CT. The most frequent variants are accessory SIJ and iliosacral complex (Figures 3A,B). These variants are sometimes associated with edematous or structural changes suspected to be mechanical in nature. Positive association between anatomical variations and degenerative changes is somewhat controversial (36, 37). To the best of the authors’ knowledge, only one study has analyzed MRI changes in morphological variants of the SIJ (38).
Accessory Sacroiliac Joints The most common variant is an accessory sacroiliac joint (3.6– 50%), which is more common in females and has a positive association with increased BMI (38, 39). Accessory SIJ is detected between the iliac and sacral articular surfaces in the posterior aspect of the joint.
It is however not certain if the accessory SIJ are congenital or acquired. In fact, degenerative ankylosis and overall structural changes may masquerade accessory SIJ.
This variation is best depicted on axial slices and is located at the level of the first or second sacral foramen. Signal intensity changes are depicted in a proportion of patients, mostly related to sclerotic or fatty changes, but rarely edematous.
Iliosacral Complex The iliosacral complex corresponds to a marked prominence of the ilium opposite a concave depression of the posterolateral sacrum (40). An iliosacral complex is present in 4% (5.8–11.7%) of individuals and is the second most common anatomical variant and seen bilaterally with slightly increased frequency in women (38). The iliosacral complex…